Skip to content

Latest commit

 

History

History
73 lines (58 loc) · 3.63 KB

README.md

File metadata and controls

73 lines (58 loc) · 3.63 KB
Raw

⚠️⚠️⚠️ Summary ⚠️⚠️⚠️

⚠️⚠️⚠️ The sklearn PyPI package is deprecated use scikit-learn instead ⚠️⚠️⚠️

See section below for some advice about how to fix the error for the main use cases.

Note that the scikit-learn PyPI package was always the official PyPI package from the start of the project: this does not change.

What changed is that the sklearn PyPI package, which is an empty package with a dependency to scikit-learn, now fails to install. See section below that explain the motivations behind this change.

How to fix the error for the main use cases

  • use pip install scikit-learn rather than pip install sklearn
  • replace sklearn by scikit-learn in your pip requirements files (requirements.txt, setup.py, setup.cfg, Pipfile, etc ...)
  • if the sklearn package is used by one of your dependencies it would be great if you take some time to track which package uses sklearn instead of scikit-learn and report it to their issue tracker
  • as a last resort, set the environment variable SKLEARN_ALLOW_DEPRECATED_SKLEARN_PACKAGE_INSTALL=True to avoid this error

If the previous advice does not cover your use case, feel free to open an issue about it.

Reason for the deprecation

sklearn package on PyPI exists to prevent malicious actors from using the sklearn package, since sklearn (the import name) and scikit-learn (the project name) are sometimes used interchangeably. scikit-learn is the actual package name and should be used with pip, e.g. for:

  • pip commands: pip install scikit-learn
  • pip requirement files (requirements.txt, setup.py, setup.cfg, Pipfile, etc ...)

At the time of writing (October 2022) sklearn downloads is about 1/5 of the scikit-learn downloads on PyPI so a lot of people are using it.

There are some edge cases with the way the PyPI sklearn package is implemented:

  • pip install sklearn==1.1.3 will say that the 1.1.3 version does not exist, which is confusing. The only available version at the time of writing of sklearn is 0.0.
  • pip uninstall sklearn will actually not uninstall scikit-learn, you can still do import sklearn afterwards
  • it can be confusing to have both sklearn and scikit-learn in the pip list output, prompting questions like "why do I have scikit-learn 1.1.3 and sklearn 0.0, and what does it even mean"?

Historical brownout schedule (from 2022-12-01 to 2023-11-31)

Starting 2023 December 1st, trying to install the sklearn PyPI package raises an error.

The table shows the historical brownout schedule that was used between 2022 December 1st and 2023 December 1st, in order to get people aware of the deprecation and give them some time to adapt. During these dates and time windows, an exception was raised if you attempted to install the sklearn package from PyPI.

Dates Window(s)
2022 December 1st - 2023 January 31st :00-:05 every hour
2023 February 1st - March 31st :00-:10 every hour
2023 April 1st - May 31st :00-:15 every hour
2023 June 1st - July 31st :00-:10 and :30-:40 every hour
2023 August 1st - September 30th :00-:15 and :30-:45 every hour
2023 October 1st - November 30th :00-:20 and :30-:50 every hour
2023 December 1st onwards always raise an exception